Greetings: First of all, I wanted to know if there was an index for this list readily availble for this list somewhere ... I see a bunch of old posts via FTP on a suburbia.net site, but I don't see an index or anything thread-based to allow be to look at/for specific items. (This may be all or in part due to the fact that I cannot use the nntpcache.org addresses for anything any more, and don't necessarily know where their current locations are.) Secondly, since I am making the post here anyway, I guess I can quickly attempt to describe my current difficulty with the system as I am trying to run it. The system we are running here is AIX 4.2 (the box it would run is closed and has no compilers, debuggers, or profilers available on it); it was compiled under 4.1/native cc/native libc's . There were a couple of glitches in the autoconfigured compile (although minor in my opinion at first site): "BAD" is #defined in a standard library on that box, and something was declared of type 'fd_set' before the compile knew what that was (and so thought it was a function, ...). My workarounds there were to rename the macro "BAD" as it appears in confused_runtime.c, and to include <sys/select.h> in extern.h . As your files mentioned, MMAP is supposed to be OK for AIX 4.x -- and your mmap tests that ran concur with that. Now, when I go to run it, and I am trying to use a local server, I get a "memory corrupt" error and the server stontaneously dies, sometimes dumping a core that dbx doesn't like. If I use a remote server, such as Microsoft's (shhh), the server runs until one tries to connect to it to get something, at which time you get the "memory corrupt" error and it dies. As this was autoconfigured, it already had the patch in it that attempts to make sure that mmcheck() is only called once and after the mmalloc, etc., as the older notes bore out for OSF. I tried running it on the 4.2 box first, and after that failed, tried to run it on the box where it was built, with the same set of results. Is this something that has already been tripped over, and if so, is there a pre-existing workaround to it? Could there be a genuine problem with the MMAP facility in 4.2, or could that error message be generated through some other quirk of execution? Thanks in advance, C@G News Administration Staff
